bama
BAMA: 모든 Mac 앱을 만들 수 있습니다
ClickHouse® 개발 및 DBA 작업을 위한 Altinity 에이전트 스킬
Altinity's agent skills for ClickHouse® development and DBA tasks
스킬Python · ShellGitHub @Altinity직접 설치운영자 미확인
써 보셨어요? 한 번만 눌러 주세요. 로그인은 필요 없어요.
개인 계정의 공개 저장소 소유자는 GitHub로 로그인해 운영자를 확인하고 소개를 직접 관리할 수 있습니다. 관리자 승인은 필요하지 않습니다.
로그인한 뒤 확인 파일을 만들어 이 저장소의 기본 브랜치에 30분 안에 커밋해 주세요. 다른 브랜치나 포크의 파일은 인정하지 않습니다. 파일을 올린 뒤 아래에서 확인하면 관리자 승인 없이 관리할 수 있습니다. 자격을 유지하려면 파일을 남겨 두세요.
로그인이 필요하면 로그인 화면으로 이동합니다. 로그인 후 이곳으로 돌아와 원하는 확인 버튼을 다시 눌러 주세요.
설치형 프로젝트는 배포 도메인 없이도 확인할 수 있습니다. 개인 저장소는 GitHub 로그인으로, 조직 저장소는 위의 저장소 확인 파일로 직접 확인해 주세요.
README 전체를 불러오는 중입니다. 아래는 저장된 발췌입니다.
This repository contains skills used for ClickHouse® DB performance and schema analysis and helper workflows.
Core Skills
• altinity-expert-clickhouse/: Modular ClickHouse diagnostic skill set. Each module is a standalone skill under altinity-expert-clickhouse/skills/ (e.g., memory, merges, replication).
• altinity-profiler-clickhouse/: Profiles a live ClickHouse cluster via MCP and generates a per-cluster analyst Skill (schema map, query patterns, engine idioms) that can be saved in claude.ai.README 전문 보기 ↗This repository contains skills used for ClickHouse® DB performance and schema analysis and helper workflows.
altinity-expert-clickhouse/: Modular ClickHouse diagnostic skill set. Each module is a standalone skill under altinity-expert-clickhouse/skills/ (e.g., memory, merges, replication).altinity-profiler-clickhouse/: Profiles a live ClickHouse cluster via MCP and generates a per-cluster analyst Skill (schema map, query patterns, engine idioms) that can be saved in claude.ai.SKILL.md.SKILL.md (e.g., modules/, scripts, prompts).Each skill group has its own release workflow and tag scheme:
| Skill group | Tag pattern | Workflow |
|---|---|---|
altinity-expert-clickhouse | expert-v* | .github/workflows/expert-skills.yaml |
altinity-profiler-clickhouse | profiler-v* | .github/workflows/profiler-skill.yaml |
On every PR / push to main, changed skills are packaged and uploaded as workflow artifacts. On a matching tag push, all skills in that group are packaged and published as GitHub Release assets.
One-liner (recommended):
npx skills add --agent claude-code Altinity/altinity-skills/altinity-expert-clickhouse/
npx skills add --agent claude-code Altinity/altinity-skills/altinity-profiler-clickhouse/
This installs all skills into ~/.claude/skills/.
Install a single skill:
npx skills add --agent claude-code Altinity/altinity-skills/altinity-expert-clickhouse/skills/altinity-expert-clickhouse-overview
Manual install:
git clone https://github.com/Altinity/altinity-skills.git
# Copy a specific skill
cp -r altinity-skills/altinity-expert-clickhouse/skills/altinity-expert-clickhouse-overview ~/.claude/skills/
# Or symlink the whole skills directory
ln -s "$(pwd)/altinity-skills/altinity-expert-clickhouse/skills" ~/.claude/skills/altinity-expert-clickhouse
Usage — invoke a skill with a slash command:
/altinity-expert-clickhouse-overview Analyze cluster health
/altinity-profiler-clickhouse Profile this cluster and generate an analyst skill
Each skill is a separate zip. Upload as many as needed.
One-liner (recommended):
npx skills add --agent codex Altinity/altinity-skills/altinity-expert-clickhouse/
npx skills add --agent codex Altinity/altinity-skills/altinity-profiler-clickhouse/
This installs all skills into ~/.codex/skills/.
Manual install:
git clone https://github.com/Altinity/altinity-skills.git
# Copy a specific skill
cp -r altinity-skills/altinity-expert-clickhouse/skills/altinity-expert-clickhouse-overview ~/.codex/skills/
# Or symlink the whole skills directory
ln -s "$(pwd)/altinity-skills/altinity-expert-clickhouse/skills" ~/.codex/skills/altinity-expert-clickhouse
Usage — invoke a skill with a dollar-sign prefix:
$altinity-expert-clickhouse-overview Analyze cluster health
$altinity-profiler-clickhouse Profile this cluster and generate an analyst skill
npx skills add --agent gemini Altinity/altinity-skills/altinity-expert-clickhouse/
npx skills add --agent gemini Altinity/altinity-skills/altinity-profiler-clickhouse/
Or manually:
git clone https://github.com/Altinity/altinity-skills.git
mkdir -p ~/.gemini/skills
ln -s "$(pwd)/altinity-skills/altinity-expert-clickhouse/skills" ~/.gemini/skills/altinity-expert-clickhouse
Usage:
use skill `altinity-expert-clickhouse-overview` Analyze cluster health
OpenCode discovers Agent Skills from a project's .opencode/skills/ directory (and .claude/skills/, .agents/skills/), following symlinks. Keep skills project-local:
git clone https://github.com/Altinity/altinity-skills.git
mkdir -p .opencode
ln -s "$(pwd)/altinity-skills/altinity-expert-clickhouse/skills" .opencode/skills
The model loads a skill with its skill tool; ask for it by name:
Use the altinity-expert-clickhouse-connection skill, then altinity-expert-clickhouse-overview, and report server health.
Tested with mid-size open models (Qwen 3.8 flash-next) at temperature: 0; see altinity-expert-clickhouse/tests/opencode/ for the evaluation harness and a ready-made read-only agent definition.
A pre-built Docker image with Claude Code, Codex CLI, and all skills is available:
docker pull ghcr.io/altinity/expert:latest
The image includes:
claude - Anthropic Claude Code CLIcodex - OpenAI Codex CLIaltinity-mcp - Altinity MCP server# Claude agent
docker run -it --rm \
-v ~/.claude:/home/bun/.claude \
ghcr.io/altinity/expert:latest \
claude --dangerously-skip-permissions -p "/altinity-expert-clickhouse-overview Analyze cluster health"
# Codex agent
docker run -it --rm \
-v ~/.codex:/home/bun/.codex \
ghcr.io/altinity/expert:latest \
codex exec --dangerously-bypass-approvals-and-sandbox "\$altinity-expert-clickhouse-overview Analyze cluster health"
A Helm chart is provided to run skills as Kubernetes Jobs in non-interactive mode.
Chart details:
altinity-experthelm/skills-agenthelm/skills-agent/README.mdFrom OCI registry (recommended):
# Use --version to pin a chart release in production.
helm install my-audit oci://ghcr.io/altinity/skills-helm-chart/altinity-expert \
--set skillName=altinity-expert-clickhouse-overview \
--set prompt="Analyze ClickHouse cluster health" \
--set-file credentials.claudeCredentials=~/.claude/.credentials.json
From local repository:
git clone https://github.com/Altinity/skills.git
cd skills
helm install my-audit ./helm/skills-agent \
--set agent=codex \
--set model=gpt-5.3-codex \
--set skillName=altinity-expert-clickhouse-overview \
--set prompt="Analyze ClickHouse cluster health and summarize top risks" \
--set-file credentials.codexAuth=~/.codex/auth.json
helm lint ./helm/skills-agent
# Render defaults
helm template my-audit ./helm/skills-agent
# Render an IRSA-enabled variant
helm template my-audit ./helm/skills-agent \
--set storeResults.enabled=true \
--set storeResults.s3Bucket=my-results-bucket \
--set storeResults.iamRoleArn=arn:aws:iam::123456789012:role/my-eks-s3-role \
--set serviceAccount.create=true
| Parameter | Description | Default |
|---|---|---|
debugMode | Create debug Pod (sleep infinity) instead of Job | false |
agent | Agent CLI: claude or codex | claude |
model | Codex model (agent=codex only) | "" |
skillName | Skill name without leading / or $ | altinity-clickhouse-expert |
prompt | Prompt passed to skill | Analyze ClickHouse cluster health |
image.repository | Docker image repository | ghcr.io/altinity/expert |
image.tag | Docker image tag | latest |
imagePullSecrets | Pull secrets for private registries | [] |
job.restartPolicy | Pod restart policy | Never |
job.backoffLimit | Job retries | 0 |
job.ttlSecondsAfterFinished | Cleanup TTL | 3600 |
job.activeDeadlineSeconds | Job timeout | 1800 |
resources | CPU/memory requests and limits | see values.yaml |
extraEnv | Additional env vars for container | [] |
clickhouse.host | ClickHouse host | localhost |
clickhouse.port | ClickHouse port | 9440 |
clickhouse.user | ClickHouse user | default |
clickhouse.password |
helm install my-debug oci://ghcr.io/altinity/skills-helm-chart/altinity-expert \
--set debugMode=true \
--set skillName=altinity-expert-clickhouse-overview \
--set prompt="Analyze ClickHouse cluster health" \
--set-file credentials.claudeCredentials=~/.claude/.credentials.json
kubectl exec -it <release-name>-altinity-expert-debug -- /bin/sh
kubectl logs <release-name>-altinity-expert-debug
IRSA mode (recommended on EKS):
helm install my-audit oci://ghcr.io/altinity/skills-helm-chart/altinity-expert \
--set skillName=altinity-expert-clickhouse-overview \
--set prompt="Analyze ClickHouse cluster health" \
--set-file credentials.claudeCredentials=~/.claude/.credentials.json \
--set storeResults.enabled=true \
--set storeResults.s3Bucket=my-results-bucket \
--set storeResults.s3Prefix=agent-results \
--set storeResults.iamRoleArn=arn:aws:iam::123456789012:role/my-eks-s3-role \
--set serviceAccount.create=true
Static AWS credentials mode:
helm install my-audit oci://ghcr.io/altinity/skills-helm-chart/altinity-expert \
--set skillName=altinity-expert-clickhouse-overview \
--set prompt="Analyze ClickHouse cluster health" \
--set-file credentials.claudeCredentials=~/.claude/.credentials.json \
--set storeResults.enabled=true \
--set storeResults.s3Bucket=my-results-bucket \
--set storeResults.awsAccessKeyId=AKIAIOSFODNN7EXAMPLE \
--set storeResults.awsSecretAccessKey=wJalrXUtnFEMI/K7MDENG/bPxRfiCYEXAMPLEKEY
When storeResults.enabled=true:
/workspace/${TIMESTAMP}/agent-execution.log.0, the full work directory uploads to s3://${S3_BUCKET}/${S3_PREFIX}/${TIMESTAMP}/.When credentials.create=false, the existing secret must contain:
codex-auth.jsonclaude-credentials.jsonclickhouse-client-config.xmlaltinity-mcp-config.yamlclickhouse-ca.crt, clickhouse-client.crt, clickhouse-client.key# Example: install using an existing secret
helm install my-audit ./helm/skills-agent \
--set credentials.create=false \
--set credentials.existingSecretName=agent-credentials \
--set skillName=altinity-expert-clickhouse-overview \
--set prompt="Run diagnostics"
kubectl get jobs -l app.kubernetes.io/instance=my-audit
kubectl logs -l app.kubernetes.io/instance=my-audit -f
kubectl get pods -l app.kubernetes.io/instance=my-audit
Helm-free alternative is available for interactive expert runs with cloudctl + kubectl.
k8s/scripts/run-expert-job.shk8s/templates/expert-job.yamlk8s/README.mdExample:
export EXPERT_CH_USER=default
export EXPERT_RUNTIME=codex
k8s/scripts/run-expert-job.sh demo prod 'my-clickhouse-password'
Altinity offers a range of services related to ClickHouse®.
© 2025-2026 Altinity Inc. All rights reserved. Altinity®, Altinity.Cloud®, and Altinity Stable® are registered trademarks of Altinity, Inc. ClickHouse® is a registered trademark of ClickHouse, Inc.; Altinity is not affiliated with or associated with ClickHouse, Inc.
| ClickHouse password |
"" |
clickhouse.tls.enabled | Enable TLS in generated configs | true |
clickhouse.tls.ca | PEM CA cert content | "" |
clickhouse.tls.cert | PEM client cert content | "" |
clickhouse.tls.key | PEM client key content | "" |
credentials.create | Create credentials/config secret from values | true |
credentials.existingSecretName | Existing secret to mount when create=false | "" |
serviceAccount.create | Create ServiceAccount | false |
serviceAccount.name | ServiceAccount name override | "" |
serviceAccount.namespace | Namespace for created ServiceAccount | "" |
serviceAccount.annotations | Extra ServiceAccount annotations | {} |
storeResults.enabled | Upload successful run logs to S3 | false |
storeResults.s3Bucket | S3 bucket name | "" |
storeResults.s3Prefix | S3 key prefix | agent-results |
storeResults.iamRoleArn | IRSA role ARN (EKS) | "" |
storeResults.awsAccessKeyId | AWS key id when not using IRSA | "" |
storeResults.awsSecretAccessKey | AWS secret when not using IRSA | "" |
storeResults.awsRegion | AWS region | us-east-1 |